This Python package accelerates OpenCV image filtering functions for the PYNQ platform. The library implements a specific set of popular image filters and feature detection algorithms. The calculation of time-consuming tasks is implemented in the Programmable Logic (PL) of the ZYNQ chip. cv2PYNQ also includes the Video-Subsystem of the base project of PYNQ. Therefore, the HDMI In and Out interfaces can be used in your application. The library calculates every filter for gray-channel images with 1080p within 16 ms if the input and output buffers are located in the contiguous memory of the chip.
Install by typing:
git clone https://github.com/wbrueckner/cv2pynq.git
cd cv2pynq/
pip3.6 install -e .
into the terminal on your Pynq-Z1 board.
The library comes with a jupyter notebook to demonstrate its usage and capabilities.
You find the notebook in the cv2PYNQ folder of your home tree after installation.
Link to YouTube Video: https://www.youtube.com/watch?v=nRxe-NqvOl8
Currently accelerated functions:
- Sobel: 3x3; 5x5
- Scharr
- Laplacian: ksize = 1; 3; 5
- blur: ksize = 3
- GaussinBlur: ksize = 3
- erode: ksize = 3
- dilate: ksize = 3
- Canny
Read the instructions in cv2PYNQ - The project behind the library.